Mojo function
schedule_group_barrier
schedule_group_barrier(mask: AMDScheduleBarrierMask, size: SIMD[int32, 1], sync_id: SIMD[int32, 1])
Controls instruction scheduling across a barrier point in AMD GPU code by creating schedule groups.
This function creates a scheduling barrier that groups instructions into sequences with custom ordering. It affects the code that precedes the barrier. The barrier ensures instructions are scheduled according to the specified group parameters.
Note: This function only has an effect on AMD GPUs. On other platforms it will raise a compile time error. The sync_id parameter allows creating multiple schedule groups that can be ordered relative to each other.
Args:
- mask (
AMDScheduleBarrierMask): A bit mask of AMDScheduleBarrierMask flags indicating which instruction types can be scheduled across this barrier. Similar to schedule_barrier masks. - size (
SIMD[int32, 1]): The number of times to repeat the instruction sequence in the schedule group. - sync_id (
SIMD[int32, 1]): A unique identifier for the group that determines the ordering of instructions within the same schedule group.
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!